# Learning to Adapt for Stereo

## TL;DR

This paper presents a novel unsupervised learning-to-adapt framework for stereo depth estimation that enhances model robustness across diverse environments, crucial for real-world applications like autonomous driving.

## Contribution

It introduces an integrated adaptation procedure within the learning objective and a confidence measure to improve online unsupervised domain adaptation for stereo models.

## Key findings

- Improved generalization to unseen environments
- Effective masking of adaptation errors
- Beneficial for real-world stereo applications

## Abstract

Real world applications of stereo depth estimation require models that are robust to dynamic variations in the environment. Even though deep learning based stereo methods are successful, they often fail to generalize to unseen variations in the environment, making them less suitable for practical applications such as autonomous driving. In this work, we introduce a "learning-to-adapt" framework that enables deep stereo methods to continuously adapt to new target domains in an unsupervised manner. Specifically, our approach incorporates the adaptation procedure into the learning objective to obtain a base set of parameters that are better suited for unsupervised online adaptation. To further improve the quality of the adaptation, we learn a confidence measure that effectively masks the errors introduced during the unsupervised adaptation. We evaluate our method on synthetic and real-world stereo datasets and our experiments evidence that learning-to-adapt is, indeed beneficial for online adaptation on vastly different domains.